Modelling the long-term fate of mercury in a lowland tidal river. II. Calibration and comparison of two models with field data.

This study presents results of the application of two-one-dimensional/dynamic-deterministic models developed to evaluate and predict mercury contamination of a lowland river, the River Yare, Norfolk, UK. As a general indication of model performance, MODEL 1 produced a prediction of the overall mass balance of mercury in the River Yare that was 6% lower than an estimate, based on measurements, and 10% lower than the prediction provided by MODEL 2. The mercury surficial sediment concentrations calculated by MODEL 1 varied from 0.46 mg kg(-1) in 1995 (Reach 3) to 3.55 mg kg(-1) in 1986 (Reaches 6 and 7), whereas the minimum mercury concentration calculated by MODEL 2 was 0.42 mg kg(-1) in 1995 (Reach 3) and the maximum was 8.45 mg kg(-1) in 1986 (Reach 5). Average mercury concentrations experimentally measured in surficial sediments ranged from 0.47 mg kg(-1) in 1995 (Reach 3) to 8.10 mg kg(-1) in 1986 (Reach 6). MODEL 2 gave an excellent prediction of mercury in fish flesh compared with the results obtained from the actual analysis of fish flesh. Both models demonstrated good ability to simulate actual values determined for all compartments, water, surficial sediments, bottom sediments, and fish.